DEV Community: Ali Farhat

OpenAI Signals a Plan for Independent Evaluators With Employee-Like Access

Ali Farhat — Sat, 12 Sep 2026 18:45:30 +0000

OpenAI has indicated that it plans to use independent evaluators with employee-like access. The company said it would follow this approach after recent internal discussions about pacing frontier AI development, and added that it will share more details soon. For now, the significance is directional: OpenAI has signaled an intention to broaden external evaluation, but it has not published an operating framework, timeline, access scope, or safeguards.

The phrase matters because external testing can differ substantially from public product use. Evaluators with access resembling that of employees could potentially assess systems under conditions that are not available to ordinary users. However, OpenAI has not defined what employee-like access would include, who would qualify as an evaluator, or how findings would be handled.

What OpenAI has signaled

OpenAI's statement contains three concrete points:

It supports the idea of independent evaluators receiving employee-like access.
It says it will adopt a similar approach.
It says further information is forthcoming, without providing a date.

That leaves the proposal at an early stage. It is not yet possible to determine whether access would apply to models, internal tools, deployment environments, safety testing processes, or another part of OpenAI's work. Nor is there information on evaluator independence, reporting procedures, confidentiality terms, or whether any resulting assessments would be made public.

Questions to watch as details emerge

The eventual design will determine whether this becomes a meaningful evaluation mechanism or a limited testing arrangement. Readers should look for clarity on several practical questions:

What systems and capabilities independent evaluators can access.
Whether evaluators can test pre-release models, deployed products, or both.
How OpenAI selects evaluators and protects their independence.
Whether evaluations produce published findings, recommendations, or only private feedback.
How the company responds when an evaluator identifies a material concern.

These details matter because the value of an external evaluation process depends on both access and accountability. Broad access without a clear route for findings to influence decisions would have a different practical effect from a process with defined reporting and response expectations.

What the signal means for businesses using AI

Businesses should not treat this statement as a new product policy or a change to the conditions under which they can use OpenAI tools. No new customer-facing capability, pricing change, or compliance requirement has been announced.

Still, the signal is relevant for organizations that are deciding how deeply to embed AI into customer support, marketing, research, internal operations, or software products. If OpenAI later publishes a structured independent-evaluation program, it could provide additional context about how the company tests advanced systems. That context may be useful when teams assess where human review, data-handling rules, and fallback processes remain necessary in their own workflows.

The immediate practical step is more modest: keep AI adoption decisions tied to the tools, documentation, agreements, and controls that are available today. A prospective evaluation framework should be monitored as it develops, rather than used as a substitute for a company's own testing of AI-assisted processes.

Businesses using OpenAI tools should turn broad safety signals into practical decisions about data handling, human review, and vendor dependence before expanding deployment. Scalevise can help assess where AI fits into day-to-day work, identify appropriate controls, and build an adoption plan around real operational needs. Scalevise's AI consultancy connects AI opportunities with workable processes. Request a consultation.

Frequently Asked Questions

What did OpenAI say about independent evaluators?

OpenAI said it plans to use independent evaluators with employee-like access and that it will share more information soon.

Has OpenAI published a timeline for the evaluator program?

No. OpenAI said it would have more to share soon, but it did not provide a date or rollout schedule.

What does employee-like access mean in this context?

OpenAI has not defined the term. It may suggest access conditions closer to those available internally, but the specific systems, permissions, and limits have not been disclosed.

Should businesses change their AI policies because of this statement?

No new customer policy or requirement has been announced. Businesses should continue to base AI use on current documentation, agreements, and their own operational controls.

Conclusion

OpenAI's statement signals a possible move toward more independent evaluation of advanced AI systems. The idea could become significant if the company defines meaningful access, evaluator independence, and a clear process for acting on findings. Until those details are published, businesses should view the announcement as an area to monitor rather than a change to current AI deployment practices.

SEO in 2026: Why Brand Signals and Entity Authority Matter Alongside Backlinks

Ali Farhat — Sat, 12 Sep 2026 18:15:30 +0000

Backlinks remain a foundational part of search engine optimization, but link volume alone is no longer an adequate authority strategy. In an AI-driven search environment, businesses also need to make their brand understandable, credible and consistently represented across the web. That means paying closer attention to brand mentions, authoritative citations, editorial coverage, entity data and topical relevance alongside traditional link building.

This is the central argument in Search Engine Land's analysis of the new SEO authority model, published April 30, 2026. The article describes a shift toward multi-signal authority, where search and AI systems can assess not only which sites link to a business, but also the context in which the business appears, its relationships to known topics and entities, and the consistency of information associated with it.

For website owners, the practical implication is clear: do not abandon backlinks. Instead, treat links as one component of a broader effort to earn recognition from relevant, trusted sources.

From link counts to multi-signal authority

Traditional off-page SEO often prioritized the number and apparent strength of backlinks. A link from a credible, relevant publication can still be valuable because it connects a site to another source and can signal editorial confidence. However, raw counts do not capture whether a business is widely recognized as a reliable source in its field.

The entity-first approach outlined by Search Engine Land puts more weight on how a brand is established across multiple signals. An entity, in this context, is a distinct, recognizable organization, person, place or concept that systems can identify and relate to other information. Consistent references to a company, its services and its expertise can help clarify what that company is and where it has authority.

AI-driven systems can interpret more than a hyperlink. The research notes that they can evaluate context, sentiment and relationships between entities. A relevant editorial mention, an accurate citation in an authoritative source, or coverage that connects a business with a well-defined subject may therefore reinforce authority even when it does not produce a conventional backlink.

Traditional link-first focus	Multi-signal authority approach
Prioritizes backlink counts and link metrics	Combines backlinks with brand, entity, reputation and PR signals
Measures whether another site links to a page	Also considers mentions, citations, context, sentiment and entity relationships
Can encourage pursuit of links as an isolated activity	Emphasizes editorial links, earned media and a credible presence across trusted sources
Focuses on individual link acquisition	Focuses on building topical authority and validating the brand as an entity

This does not mean every unlinked mention has the same value, or that links have ceased to matter. Rather, the available evidence supports a more complete view: quality, relevance and corroboration matter more than accumulating links without regard to where they come from or what they say about the brand.

How businesses can adapt their off-page SEO

A practical strategy starts with replacing isolated link targets with a clearer authority objective. Ask which topics the business wants to be associated with, which trusted sources serve those audiences, and whether the company is represented accurately when it is mentioned elsewhere.

The following actions align with the multi-signal model described in the research:

Pursue editorially earned coverage that is relevant to the business's expertise, customers or market rather than seeking links for their own sake.
Keep core business information consistent across citations and authoritative references, including the company name and other entity details used publicly.
Create genuinely useful expert material that gives journalists, publishers and industry sources a reason to cite or reference the business.
Strengthen topical relevance by concentrating public-facing content and outreach around areas where the company can credibly demonstrate knowledge.
Monitor brand presence beyond backlinks, including where the business is cited, how it is described and whether its associations support its intended positioning.

The shift also changes how teams should evaluate digital PR. A campaign should not be judged only by referral traffic or the number of followed links it generates. Relevant media coverage and accurate brand citations can contribute to a more coherent footprint across the sources that search and AI systems encounter.

This calls for restraint as well as activity. Chasing broad exposure with little topical connection may not strengthen a company's authority in the areas that matter most. The better question is whether a placement, citation or mention helps establish a clear, credible relationship between the brand and the expertise it wants to be known for.

Search Engine Land also points to possible future measures such as Share of Model, or SoM, and a growing role for brands as primary information sources. These are forward-looking observations rather than established requirements. Still, they reinforce the value of building durable, proprietary expertise and ensuring that trusted third parties can accurately identify and reference it.

For businesses with limited time and budget, this is useful because it favors focus over scale. A smaller number of high-quality, relevant signals can be more strategically meaningful than a large volume of weak or disconnected links. The goal is to make every off-page activity contribute to a recognizable and trustworthy digital identity.

As AI-generated answers become another way people discover companies and solutions, it is increasingly important to know whether a brand is visible and accurately represented in those results. Scalevise can help you assess where your business appears, identify gaps in its AI search presence and prioritize practical improvements through its AI Visibility and GEO Checker. Start an AI visibility scan to turn scattered brand signals into a clearer search strategy.

Frequently Asked Questions

Are backlinks still important for SEO in 2026?

Yes. Backlinks remain a foundational SEO signal, particularly when they are relevant and editorially earned. The emerging approach is to use them alongside brand mentions, citations, entity consistency, reputation signals and topical authority.

What are entity signals in SEO?

Entity signals are information that helps systems identify a business as a distinct organization and understand its relationships to topics, sources and other entities. Consistent business information, authoritative citations and relevant editorial references can reinforce those signals.

Do unlinked brand mentions matter for AI-driven search?

The research indicates that AI systems can interpret brand mentions in context, including sentiment and entity relationships. An unlinked mention is not the same as a backlink, but relevant and credible mentions can contribute to broader authority signals.

What should a business prioritize instead of chasing more links?

Prioritize relevant editorial links, earned media, accurate citations, consistent entity information and content that demonstrates real expertise in the topics the business wants to own.

Conclusion

SEO authority is becoming broader than a backlink profile. Businesses that continue to earn high-quality links while building credible brand, entity and reputation signals will be better aligned with the multi-signal authority model emerging in AI-driven search. The most durable approach is not to chase link volume, but to build consistent recognition from relevant and trusted sources.

Dario Amodei’s AI Policy Agenda Calls for Pacing Frontier Development

Ali Farhat — Sat, 12 Sep 2026 16:15:30 +0000

Dario Amodei has published a policy essay arguing that the pace of frontier AI development is outstripping the ability of public policy to respond. In Policy on the AI Exponential, Amodei sets out a five-area agenda intended to pace the development and deployment of the most advanced AI systems while preserving their potential benefits.

The essay matters because it moves beyond broad calls for AI safety into specific policy areas: frontier-model testing, employment effects, civil liberties, the management of downstream technologies, and international coordination among democracies. For companies adopting AI tools, the immediate message is not that everyday use of AI should stop. It is that the rules, safeguards, and economic conditions around powerful models may change quickly as policymakers respond to faster technical progress.

Amodei’s official Policy on the AI Exponential essay frames the issue as a gap between accelerating AI capabilities and institutions designed for slower-moving technological change. Anthropic also intends to release legislative proposals on frontier-model testing and a policy framework addressing job displacement, according to the essay.

What Amodei’s policy agenda proposes

The essay presents five connected policy areas, rather than treating model safety as an isolated technical issue. Its central premise is that society needs greater capacity to assess and govern frontier systems before their most consequential uses become routine.

The five areas of the agenda

Policy area	Focus in the essay	Why it matters
Regulation and public safety	An FAA-like model, mandatory third-party testing for frontier models, and government authority to block deployments that fail safety checks.	It would create a more formal process for evaluating the most capable AI systems before deployment.
Macroeconomics and tax policy	Data collection on AI-driven displacement and interventions intended to support employment.	It recognizes that AI’s effects extend beyond software capability to jobs and economic policy.
Accelerating positive impact	Managing downstream technologies without unduly slowing beneficial progress.	It places AI’s potential benefits alongside the need to manage associated risks.
The state and civil liberties	Governance, civil rights, and accountability.	It addresses how AI use can affect the relationship between institutions and individuals.
Leadership by democracies	A coalition to coordinate AI policy, supply chains, and defense against adversaries.	It treats AI development as an international strategic issue as well as a domestic policy question.

The proposed FAA-like approach is one of the most concrete elements. Amodei argues for mandatory independent testing of frontier models and for government powers to prevent deployment when systems do not pass safety checks. The essay does not present this as a blanket restriction on all AI software. Its focus is frontier models, the systems at the leading edge of capability.

That distinction is important. A business using an established AI tool for drafting, search, customer support, or internal analysis is not the same as a developer releasing a new frontier model. Still, changes to rules for model developers can affect which models and features become available, how providers document safety, and what information customers may expect from vendors.

What the agenda could mean for AI adoption

For most organizations, the essay is primarily a signal that AI adoption cannot be separated completely from the wider policy environment. The proposed framework remains a policy agenda, not a set of enacted requirements described in the essay. But its emphasis on testing, accountability, displacement data, and civil liberties identifies issues that may become more prominent in product choices and public debate.

Keep adoption practical and reviewable

The essay’s strongest business implication is the value of purposeful AI adoption. Teams can focus on well-defined work where AI can assist people, improve a process, or reduce repetitive effort, rather than treating access to a more capable model as an end in itself.

A practical approach includes:

Identifying the specific task an AI tool is meant to improve.
Keeping human review where incorrect output could materially affect customers, employees, or decisions.
Recording what data is provided to an AI service and which team is responsible for the workflow.
Reviewing whether a provider’s product changes alter the workflow’s usefulness or reliability.
Tracking policy developments that could affect the availability or use of frontier-model features.

These are operational choices, not a claim that the essay creates new obligations for every AI user. They help businesses maintain clarity as AI products and the policy debate evolve. They also make it easier to assess whether a tool is delivering a useful outcome rather than simply adding another disconnected application to the stack.

The employment section deserves particular attention. Amodei calls for better data on AI-driven displacement and pro-employment interventions. The essay does not establish a quantified effect on any role or industry. Its significance is that workforce effects are treated as a policy question requiring measurement, not an assumption that can be resolved by either optimism or alarm.

For managers, that supports a balanced implementation question: where can AI augment existing work, and where does automation change a task enough to require redesign, training, or clearer accountability? The answer will vary by workflow. The essay provides no universal prescription for individual companies, but it makes the case that the cumulative economic effects of AI deployment should be actively examined.

The civil-liberties and democratic-leadership sections broaden the agenda further. They position frontier AI as a matter involving public accountability, rights, supply chains, and geopolitical coordination. Businesses may not directly participate in those policy decisions, but they can expect such debates to shape the environment in which AI providers operate.

If your team is moving from isolated AI experiments to repeatable operational use, Scalevise can help turn promising tools into workflows with clear goals, sensible human oversight, and measurable business value. Our AI consultancy services help businesses prioritize practical use cases and plan implementation without losing sight of changing technology and policy conditions. Request a consultation to map the AI opportunities that fit your operations.

Frequently Asked Questions

What is Dario Amodei’s Policy on the AI Exponential?

It is a June 2026 policy essay arguing that frontier AI progress is accelerating faster than policy. It proposes a five-area agenda covering safety regulation, economic effects, beneficial uses, civil liberties, and democratic coordination.

Does the essay call for stopping AI development?

No. The essay argues for pacing frontier AI development and deployment. It also emphasizes managing downstream technologies without unduly slowing beneficial progress.

What frontier AI safety measures does the essay propose?

Amodei proposes an FAA-like regulatory model, mandatory third-party testing for frontier models, and government authority to block deployments that fail safety checks.

What should businesses take from the policy agenda?

The agenda is not a new set of business requirements. It suggests that businesses should adopt AI for clear use cases, retain appropriate human review, understand their workflows, and monitor changes in AI products and policy.

Conclusion

Amodei’s essay is a detailed case for matching frontier AI’s accelerating development with stronger public capacity to test, pace, and govern it. Its five-area agenda does not call for abandoning beneficial AI use. Instead, it argues that safety, economic impacts, rights, and international coordination must develop alongside the technology. For businesses, the practical response is disciplined adoption focused on useful, understandable workflows as the policy landscape develops.

ChatGPT Ads Restrict Rival Image and Audio AI Tools as OpenAI Sets New Guardrails

Ali Farhat — Sat, 12 Sep 2026 14:15:30 +0000

OpenAI has formalized advertising guardrails for ChatGPT that can restrict campaigns promoting rival image-generation and audio-generation tools. The change gives OpenAI explicit discretion to decline, limit, or remove advertisers and ad content that conflict with its advertising principles, business interests, or competitive position. For marketers of AI creative tools, the practical result is that ChatGPT may not be a viable paid acquisition channel for certain competing products.

The restriction matters because OpenAI's advertising program is still evolving. OpenAI announced initial testing in January 2026, with later expansion to partner ecosystems and measurement capabilities during the year. As the program develops, advertisers cannot assume that access to an emerging AI audience means every product category will be accepted.

What OpenAI's ChatGPT ad policy now allows

OpenAI's official advertising policies state that it reserves the right to decline, restrict, or remove advertisers and refuse advertising content. The stated grounds include conflicts with OpenAI's advertising principles, business interests, or competitive position. That language creates a formal policy basis for limiting ads from direct competitors.

Industry reporting indicates the policy is being applied to rival image-generation and voice-generation tools. The available information does not describe a blanket ban on all AI-related advertising. Instead, it establishes that competitive overlap can affect whether a campaign is accepted or remains eligible in ChatGPT.

A useful distinction is needed here. Reports indicate that video ads remain eligible, while image- and audio-centric advertising faces restrictions. That describes the reported eligibility of ad formats and categories, not a complete public list of every product type that OpenAI will approve. The policy's broad discretion means an advertiser's final eligibility can depend on OpenAI's assessment.

Advertising area	Position indicated by the research	What marketers should take from it
Rival image-generation tools	Restricted in ChatGPT advertising	Do not treat ChatGPT as an assured paid channel for these campaigns.
Rival audio and voice-generation tools	Restricted in ChatGPT advertising	Build acquisition plans that do not depend on ChatGPT ad approval.
Video ads	Reports indicate they remain eligible	Eligibility should still be checked against current policy and campaign review.
Other advertisers and ad content	Subject to OpenAI's discretion	Approval is not guaranteed simply because a category is not named in reporting.

A platform policy, not a universal ad-market rule

This is a ChatGPT advertising policy decision, not evidence that rival AI tools cannot advertise elsewhere. Search, social, display, and other advertising platforms operate under their own policies and review systems. The immediate change is narrower: a company selling an image or voice AI product may lose, or be unable to access, a potential paid placement within ChatGPT.

That distinction is important for budget planning. Businesses should not overreact by treating the development as a broad restriction on AI marketing. They should, however, avoid allocating spend to an emerging channel before confirming that the product category, creative, and destination are eligible.

What marketers should do next

For companies promoting creative AI products, the policy makes channel diversification more important. A paid-media plan that depends on one new AI platform has more exposure to policy changes than one that can shift demand generation across established channels and owned audiences.

Practical steps include:

Review whether your product directly overlaps with image-generation or audio-generation capabilities that OpenAI may regard as competitive.
Confirm current campaign eligibility before committing creative-production or media budget to ChatGPT placements.
Keep campaign assets and measurement plans usable across other approved channels rather than building around a single ad pilot.
Separate paid acquisition planning from ChatGPT-based operational workflows, since ad eligibility and use of ChatGPT in business processes are different questions.

The last point is especially relevant for teams already using ChatGPT in research, content operations, customer support, or internal workflows. The ad policy does not change those uses. It changes the commercial terms under which certain companies may seek to advertise inside ChatGPT.

OpenAI's staged advertising rollout also means the rules may continue to develop alongside formats, partner distribution, and measurement features. Marketers should therefore treat the current restriction as a planning constraint, not as a final map of the ChatGPT ad ecosystem.

If ChatGPT and other AI assistants are becoming part of how customers discover businesses, paid placement is only one part of the opportunity. Scalevise can help you understand how your brand appears in AI-driven answers and identify practical visibility gaps through its AI Visibility and GEO Checker. This gives your team a clearer basis for prioritizing content, brand signals, and channel investment when ad access or policies change. Start an AI Visibility scan.

Frequently Asked Questions

What does OpenAI's ChatGPT ads policy restrict?

OpenAI's policy gives it discretion to decline, restrict, or remove advertisers and ad content that conflict with its advertising principles, business interests, or competitive position. Reporting indicates this has restricted ads for rival image-generation and audio or voice-generation tools.

Are all AI tool ads banned from ChatGPT?

No. The supplied research does not establish a blanket ban on AI tool advertising. It identifies restrictions affecting competing image and audio tools, while OpenAI retains broad discretion over advertiser and content eligibility.

Are video ads still allowed in ChatGPT?

Reports indicate that video ads remain eligible. However, OpenAI's policy allows it to restrict or refuse advertisers and ad content, so businesses should confirm eligibility before committing campaign budget.

Does this change how businesses can use ChatGPT in their workflows?

No. The policy change concerns advertising within ChatGPT. It does not change the research-supported use of ChatGPT in business workflows such as research, content operations, customer support, or internal work.

Conclusion

OpenAI's updated ad policies make competitive position a clear factor in ChatGPT advertising eligibility. For rival image and audio AI providers, that can remove a potential paid channel and make early policy checks essential. The wider lesson for marketers is straightforward: as AI advertising develops, channel access may be shaped as much by platform rules as by audience demand.

TrustED Moves Privacy Preserving Data Sharing Into Real World European Pilots

Ali Farhat — Sat, 12 Sep 2026 12:30:30 +0000

The EU-funded TrustED project is moving privacy-preserving data sharing from technical design into real-world validation. Its two pilots combine digital identity, user-controlled credentials and privacy-enhancing technologies to help organisations collaborate and analyse information without routinely exposing sensitive underlying data. For European businesses, the work offers a practical view of how future data spaces could support cooperation while preserving privacy and control.

TrustED stands for Enabling Trustworthy European Data Spaces through Self-Sovereign Identity and Privacy Preserving Technologies. The project is developing a secure, interoperable identity framework designed to work with the European Digital Identity Wallet. Its aim is not simply to transfer data more securely. It is to let people and organisations prove relevant facts, manage verifiable credentials and participate in data-sharing arrangements with greater control over what is disclosed.

The European Commission's CORDIS feature on TrustED highlights the project as part of Europe's effort to create secure, human-centred digital infrastructure. TrustED has also been featured as a CORDIS Project of the Month, reflecting its place in the wider European data-sharing and digital-identity landscape.

From identity design to two applied pilots

TrustED's current work is most concrete in two pilots. One focuses on collaboration between volunteers and non-governmental organisations. The other examines clinical data analysis across institutions. They address different sectors, but both test a shared proposition: useful collaboration should not require every participant to hand over all of their underlying sensitive information.

Area	TrustED pilot	What it is evaluating
Volunteering and NGOs	A self-sovereign identity wallet for managing verifiable credentials, linked to a data-space dashboard for volunteering opportunities.	Cross-organisation collaboration and user control over credentials.
Clinical data analysis	A federated learning setup for analysing clinical data without sharing raw data.	Governance, clinical workflows and trust in cross-institution collaboration.

In the volunteering pilot, a wallet gives volunteers and NGOs a way to manage verifiable credentials. The related data-space dashboard is intended to support access to volunteering opportunities across organisations. The pilot therefore tests a common data-sharing obstacle: how several organisations can rely on trustworthy credentials without creating unnecessary copies of personal information in each system.

The clinical pilot uses federated learning, an approach in which data analysis can be performed while raw data remains with the institution that holds it. TrustED is evaluating this setup alongside governance, clinical workflows and trust. That scope matters because privacy-preserving technology has to fit real operational processes, not only work in a technical demonstration.

The privacy technologies behind the pilots

TrustED brings several privacy-enhancing technologies, often called PETs, together with self-sovereign identity and the European Digital Identity Wallet. The official project materials cite:

Federated learning, which supports analysis without centrally sharing raw data.
Zero-knowledge proofs, which can enable a party to demonstrate a fact without disclosing more information than necessary.
Homomorphic encryption, a technique intended to enable computation on encrypted information.
Differential privacy, which is designed to reduce the risk of identifying individuals in analytical outputs.

These technologies serve different purposes. Federated learning addresses where data is analysed, while cryptographic and statistical techniques can limit what is revealed in verification and analysis. TrustED's contribution is the attempt to make these elements work together in a data-space setting where digital identity, consent, credentials and collaboration need to interact.

What TrustED could mean for business data strategies

TrustED does not establish a finished, generally available business product or a universal data-sharing standard. The project is still advancing through implementation and pilot evaluation. Its relevance is that it tests building blocks that may be useful where organisations need to collaborate but cannot, or should not, freely exchange raw personal or commercially sensitive data.

For example, a company working with partners may need to verify that an individual has a valid role, qualification or permission. A verifiable credential could potentially reduce the need for repeated manual checks and avoid sharing a full record when only one attribute needs to be confirmed. Similarly, organisations with complementary datasets may be interested in joint analysis, but unable to centralise those datasets because of sensitivity or trust concerns. Federated approaches point to a different model: take analysis to the data rather than automatically moving data to a shared repository.

The pilots also underline that privacy is not only an encryption question. A workable arrangement needs clear identity controls, credible credentials, compatible systems and workflows that participants can actually use. Businesses considering participation in European data spaces should therefore assess more than the volume of data they possess. They should identify which data can be shared, which facts need to be verified, who controls access and whether a proposed collaboration can deliver value without disclosing raw records.

What happens next will depend on the results of TrustED's pilots and its continuing work on integration with Europe's digital-identity and data-space infrastructure. The confirmed progress is meaningful, but the supplied project information does not specify a commercial rollout date, broad market availability or pricing. Readers should treat TrustED as an active validation effort, rather than assume that all of its capabilities are ready for routine deployment.

For businesses exploring secure partner data flows, the lessons from TrustED are immediate: define the minimum information a process needs, then design identity and access around that requirement. Scalevise can help translate that principle into an adoption roadmap, assess suitable AI and privacy-focused use cases, and connect practical technology choices to existing operations. Explore Scalevise's AI consultancy to request a consultation on a practical data and AI strategy.

Frequently Asked Questions

What is the TrustED project?

TrustED is an EU-funded project developing a secure, interoperable digital identity framework for European data spaces. It combines self-sovereign identity with privacy-enhancing technologies and is designed to work with the European Digital Identity Wallet.

What pilots has TrustED demonstrated?

TrustED has demonstrated a self-sovereign identity wallet for volunteers and NGOs, linked to a volunteering data-space dashboard, and a federated learning setup for secure clinical data analysis across institutions.

How does federated learning protect sensitive data?

Federated learning enables data analysis without sharing raw data centrally. In TrustED's clinical pilot, it is being evaluated for cross-institution collaboration alongside governance, clinical workflows and trust.

Is TrustED a commercially available product?

The supplied project information describes TrustED as an active research and validation project. It does not specify broad commercial availability, a rollout date or pricing.

Why should businesses follow TrustED?

TrustED is testing approaches that could help organisations verify credentials and conduct collaborative analysis while limiting exposure of personal or sensitive data. Its results may be relevant to organisations planning participation in European data spaces.

Conclusion

TrustED's significance lies in its shift from architecture to applied testing. By validating identity wallets and privacy-preserving analytics in volunteering and clinical settings, the project is examining whether European data spaces can support useful collaboration without making broad raw-data sharing the default. The pilots will not answer every implementation question, but they provide a concrete reference point for organisations building more privacy-conscious data strategies.

Google Ads Is Shifting Toward Longer Queries: How Advertisers Should Reallocate Spend

Ali Farhat — Sat, 12 Sep 2026 02:15:30 +0000

Google Ads search behavior is becoming more specific. Search Engine Land reported that one- and two-word queries fell from 42% to 24% of impression share, while three- and four-word queries rose from 33% to 48%. For advertisers, the change makes mid-tail keyword coverage, intent-led bidding, and more precise ad messaging increasingly important.

The pattern also appears in Google’s AI Overview results. Adthena’s analysis of roughly 29.1 million queries across more than 10 industries in the US, EMEA, and APAC, covering November 2025 through March 2026, found that three- and four-word searches were the dominant range for AI Overview appearances. The company’s analysis of AI Overview visibility frames this mid-tail range as a key area advertisers can influence through more deliberate search strategy.

Why mid-tail queries matter now

A three- or four-word query often carries more usable context than a broad one- or two-word search. It can signal the type of product, service, problem, location, or stage of consideration that matters to the searcher. That does not mean every longer phrase converts better, or that short keywords have lost their value. It does mean advertisers should not let historically broad, high-volume terms absorb budget without testing whether more explicit intent is now producing stronger visibility and business outcomes.

The available data covers two related but distinct signals. The Search Engine Land figures describe a shift in impression share for non-AI Overview terms. Adthena’s research focuses on the query lengths associated with AI Overview appearances. Together, they suggest that more specific searches are becoming a more important part of the search landscape, but they do not provide a universal conversion benchmark for every advertiser, industry, or market.

Query range	Reported impression-share change	AI Overview signal from Adthena
1-2 words	Fell from 42% to 24%	Not identified as the dominant range
3-4 words	Rose from 33% to 48%	Dominant range for AI Overview appearances
5-6 words and 7+ words	No overall impression-share figure supplied	Gaining some conversion share in certain pockets

Adthena reported substantial regional variation in the three- to four-word share of AI Overview appearances: 54.9% in the US, 73.1% in Australia, and 74.6% in Asia. Related UK coverage placed AI Overviews in more than 17% of searches. Those differences matter. A company should assess its own markets and search categories rather than assume that a global pattern will appear at the same speed in every campaign.

Reassess keyword coverage before expanding spend

The practical first step is a keyword audit centered on query length and intent. Separate broad one- and two-word terms from three- and four-word phrases, then identify where the latter reveal a clearer need that existing ads or landing pages do not address. The objective is not simply to add words to a keyword list. It is to capture the meaningful qualifiers people increasingly use.

A useful review should cover:

Mid-tail gaps where a product category, use case, service type, or local qualifier is absent from current coverage.
Longer-tail variants that may reveal narrower needs and deserve controlled testing rather than automatic scale.
Search terms with unclear intent that currently consume budget through very general queries.
Overlap between ad groups so more specific terms receive suitable messaging and are not obscured by broad targeting.

Match-type strategy should be part of that review. The research does not prescribe a single match type, but it does support revisiting how broad, phrase, and exact targeting are used when search demand is moving toward more detailed language. Advertisers need enough breadth to discover emerging variations, while retaining enough control to evaluate whether those variations reflect commercially relevant intent.

Align bids and creative with explicit intent

A growing share of three- and four-word searches is a reason to examine budget allocation, not a reason to impose a blanket bid increase. Compare spend, impression share, and the outcomes that matter to the business across short-tail, mid-tail, and longer-tail groups. If mid-tail terms are producing valuable engagement or conversions, they may warrant a larger share of the testing budget than broad category terms.

Ad copy also needs to meet the searcher at the level of specificity visible in the query. Where a phrase indicates a particular use case or problem, the ad should directly address that need rather than repeat a generic category label. This is especially relevant as AI Overviews change how visibility is distributed on search results pages. A precise message can help an advertiser remain relevant when a user has already expressed more of their intent before seeing an ad.

Longer search terms should still be treated carefully. Adthena’s findings indicate that five- to six-word and seven-plus-word phrases are gaining some conversion share in certain areas, but the central pattern is the rise of the three- to four-word mid-tail. Test longer phrases where they map to real offerings, then judge them against the company’s own performance data instead of assuming query length alone determines value.

Measurement remains an open issue

AI Overview reporting and inventory visibility are still evolving, which limits the precision of any universal playbook. Advertisers cannot assume that an AI Overview appearance maps directly to a particular paid outcome, and regional penetration varies. The more durable response is to establish a repeatable way to monitor query-length trends, allocate test budgets, and review creative relevance.

More specific queries can improve the quality of decisions only if campaigns are structured to surface those differences. Group terms by intent and length, measure their performance consistently, and avoid treating all search traffic in a category as interchangeable. That approach gives advertisers a clearer basis for deciding where to expand, where to restrain spend, and which messages require revision.

More specific searches and AI-driven search experiences make broad visibility alone less reliable as a planning signal. Scalevise can help businesses assess whether their brand and offers appear in AI-driven search experiences, identify intent gaps, and prioritize the content worth improving. Start an AI Visibility scan with Scalevise's GEO Checker.

Frequently Asked Questions

What is changing in Google Ads query behavior?

Search Engine Land reported that one- and two-word queries fell from 42% to 24% of impression share, while three- and four-word queries rose from 33% to 48%.

Does AI Overview data show the same mid-tail trend?

Yes. Adthena found that three- and four-word queries were the dominant range for AI Overview appearances, representing 54.9% in the US, 73.1% in Australia, and 74.6% in Asia.

Should advertisers stop using short-tail Google Ads keywords?

No. The data supports reassessing short-tail budget allocation and testing more specific terms, not abandoning broad category keywords. Results can vary by market, industry, and search intent.

How should advertisers adapt their campaigns?

Audit keyword coverage for three- and four-word intent gaps, test longer variants, review match-type strategy, compare budget performance by query length, and write ads that answer the specific need expressed in the search.

Conclusion

The shift toward three- and four-word searches gives Google Ads advertisers a clear reason to revisit campaigns built around broad category terms. Mid-tail queries are gaining impression share and dominate the AI Overview appearances measured by Adthena. The most practical response is disciplined testing: expand relevant intent coverage, evaluate bids by query group, and make ad messaging match the specificity of the search.

GPT-5.6 Sol Runs Quantum Chip Calibration at MIT Through Codex

Ali Farhat — Fri, 11 Sep 2026 22:30:30 +0000

OpenAI says its GPT-5.6 Sol model, harnessed through Codex, helped run routine quantum computing experiments in MIT's Engineering Quantum Systems Group. In the demonstration, graduate student Beatriz Yankelevich used an agent to coordinate measurements on a superconducting six-qubit chip, shifting a substantial part of a repetitive calibration workflow from manual execution to AI-assisted operation.

The significance is not that an AI system independently replaced a physicist. It is that the system was applied to a defined, hardware-connected scientific process: selecting measurement parameters, operating laboratory equipment, analysing results and determining the next step. As described in OpenAI's case study on Codex and quantum computing experiments, the result gave the researcher more time for experiment design, data analysis and planning.

For businesses watching AI move beyond chat and content generation, the MIT work is a useful example of a more practical direction. AI agents can become valuable when they are connected to real tools and given bounded, repeatable workflows with clear outputs, while people retain responsibility for ambiguous cases and higher-level decisions.

From routine measurements to an agent-led workflow

Quantum-chip calibration involves interdependent measurements that need to be performed repeatedly. OpenAI says the Codex-enabled agent handled an end-to-end set of clearly defined tasks, including identifying qubit transition frequencies, calibrating control and readout pulses, and estimating coherence.

Those steps matter because calibration is not a single isolated command. Measurements affect later choices, and the workflow requires the system to interpret data before proceeding. In this deployment, GPT-5.6 Sol selected parameters, ran the relevant measurements through the lab hardware, assessed the results and chose subsequent actions with minimal human intervention.

The work took place in MIT's Engineering Quantum Systems Group, or EQuS, where Yankelevich is a graduate student. The case study therefore describes a real laboratory setting rather than a purely simulated benchmark. It also narrows the claim appropriately: the agent automated a substantial portion of a routine workflow, not the full scientific process of developing quantum computing research.

Workflow element	Routine manual approach	MIT demonstration using GPT-5.6 Sol through Codex
Measurement parameters	Researcher performs repeated calibration choices	Agent selected measurement parameters
Laboratory execution	Researcher performs routine measurements	Agent operated lab hardware for the workflow
Results and next actions	Researcher analyses results and plans follow-up steps	Agent analysed results and decided subsequent steps
Ambiguous data	Researcher judgment is required	Human guidance may still be required

What the demonstration does and does not show

The strongest conclusion from the case study is that an agent can coordinate a structured experimental sequence, rather than merely answer questions about one. That is a meaningful advance for laboratory automation because the work combines tool use, data interpretation and repeated decision-making inside a defined process.

It does not establish that AI can reliably manage every experimental condition without oversight. OpenAI explicitly notes that noisy or ambiguous data may still need human guidance. That limitation is important in quantum research, where measurements can be difficult to interpret and calibration decisions can have consequences for later experiments.

The practical model is therefore automation with an escalation path. The agent can take on routine, well-specified work. The researcher can intervene when data quality, exceptions or scientific judgment require it. This division of labour is also more realistic for companies than a promise of fully autonomous operations.

Why this matters outside a quantum lab

Most businesses do not operate a superconducting quantum chip, but many have workflows with the same basic structure: gather data, apply a known procedure, inspect the result and determine the next action. The relevant lesson is not to deploy an AI agent everywhere. It is to identify work that is repetitive, measurable and connected to approved systems.

Potential candidates include:

recurring simulations or test runs with established parameters
quality-control checks that produce structured measurements
data-processing workflows that need routine follow-up actions
operational tasks that require moving between software tools and interpreting defined outputs

The MIT example also highlights the prerequisites. An agent needs access to the relevant tools, an understandable workflow and boundaries for when it should stop or ask for help. In the quantum demonstration, the tasks were concrete: locate transition frequencies, calibrate pulses and estimate coherence. Businesses should seek a similarly specific starting point rather than beginning with an open-ended request to automate an entire department.

That approach can help teams concentrate scarce expert time on exceptions, planning and higher-value analysis. It can also expose process weaknesses. If a workflow cannot be described clearly enough for an agent to execute safely, it may need better documentation, cleaner data or more consistent operating procedures before automation will deliver dependable results.

For companies exploring agent-led operations, Scalevise can help turn a repeatable process into a controlled implementation that connects AI to the tools your team already uses. Our AI workflow automation service focuses on reducing manual handoffs, defining practical human review points and building workflows around measurable business outcomes. Start by identifying one high-volume task with clear inputs and decisions, then discuss an AI automation project with Scalevise.

Frequently Asked Questions

What did GPT-5.6 Sol do in MIT's quantum computing experiment?

OpenAI says the Codex-enabled agent selected measurement parameters, operated lab hardware, analysed results and chose subsequent steps for routine measurements on a superconducting six-qubit chip.

Did GPT-5.6 Sol fully automate quantum computing research at MIT?

No. The demonstration automated a substantial portion of a defined calibration workflow. OpenAI notes that noisy or ambiguous data may still require human guidance.

Which quantum computing tasks did the agent handle?

OpenAI says the workflow included identifying qubit transition frequencies, calibrating control and readout pulses, and estimating coherence.

What is the business lesson from this MIT demonstration?

The case suggests that AI agents are most useful for bounded, repeatable workflows that combine tool access, structured data and clear points for human intervention.

Conclusion

The MIT demonstration shows a concrete application of GPT-5.6 Sol through Codex: coordinating routine quantum-chip calibration tasks in a working research environment. Its wider relevance lies in the workflow design. AI agents can take on meaningful operational sequences when the tasks, systems and escalation points are clearly defined, while expert people remain essential for uncertainty and scientific judgment.

Claude CRO Audit Workflow: Faster Data Triage With Human Evidence Validation

Ali Farhat — Fri, 11 Sep 2026 22:15:30 +0000

Claude can speed up the early stages of a conversion-rate-optimization audit, but it should not be treated as the system that decides what is true or what to test. A documented workflow published by Search Engine Land on September 10, 2026, sets out a practical role for Claude: triaging analytics, organizing evidence, and drafting structured findings while people retain responsibility for data quality, business context, validation, and prioritization.

The central lesson from Search Engine Land's Claude CRO audit workflow is that useful AI assistance starts with bounded tasks. Asking an LLM to simply “run a CRO audit” risks producing broad recommendations that sound polished but are weakly grounded. Giving Claude a defined dataset, an explicit output format, and a specific validation question makes its contribution more inspectable.

For marketing teams, that distinction matters. CRO work often begins with a large set of landing pages, segments, traffic changes, and funnel metrics. Sorting that material is repetitive and time-consuming. Claude can help narrow the field of investigation, but a faster shortlist is not the same as a validated explanation for why a page is underperforming.

A controlled role for Claude in CRO audits

The workflow begins by building an evidence pack that separates three inputs: performance data, observations about the page or journey, and relevant business context. Keeping these categories distinct helps prevent an AI-generated narrative from blending a measured fact with an assumption about user behavior.

Claude accepts common working formats, including CSV files, PDFs, DOCX documents, JSON, and HTML. Files can also be stored in a Project where appropriate. This gives teams a way to work from exported analytics reports and supporting materials without treating the model as a replacement for their analytics stack.

Start with data triage, not recommendations

The recommended first task is data triage. For example, a team can ask Claude to review an attached GA4 landing-page report and return a defined set of fields: the page and segment, sessions, conversions, period-over-period changes, evidence references, possible explanations, and the next validation step.

That structure turns Claude into an assistant for finding areas worth investigating. It also makes the output easier to review because each possible explanation should point back to the supplied evidence and specify what needs checking next.

A practical triage output can help teams identify pages with meaningful shifts before they spend time on deeper qualitative review. It should not, however, be used as proof that a particular design issue, message problem, or audience mismatch caused the change. Claude cannot determine whether the source data is trustworthy, complete, or interpreted correctly for the business.

Choose static exports or live data access deliberately

The workflow distinguishes between static exports and connections to live sources through the Model Context Protocol (MCP). Static files are appropriate when a team needs a bounded snapshot for a defined analysis. MCP can be useful when Claude needs controlled access to approved sources such as GA4, Google Search Console, a CRM, or a data warehouse.

Approach	Best fit in the workflow	Control considerations
Exported files	A defined, point-in-time evidence pack for a specific audit task	Keep the dataset bounded and preserve the source material used for review
MCP-connected sources	Approved live access to sources such as GA4, Search Console, CRM systems, or data warehouses	Use read-only access, least-privilege permissions, data minimization, and an audit trail

The choice is not simply about convenience. Live connections can reduce manual export work, but they increase the importance of access design. The documented guardrails emphasize read-only permissions, granting only the minimum access needed, reducing the data passed to the model, and retaining a clear record of how information was used. These controls help a team make AI-assisted analysis more repeatable without giving an assistant unnecessary reach into business systems.

Turn observations into testable findings

Once triage has identified a smaller set of pages or audiences, Claude can help organize the resulting findings into a prioritization framework. The suggested structure includes the page or audience, supporting evidence, expected behavior change, hypothesis, primary metric, guardrails, confidence, and implementation effort.

This is valuable because it separates an idea from a test plan. A finding with a stated primary metric and guardrails is easier to compare with other opportunities than a generic recommendation to “improve the page.” It can also expose missing information. If the team cannot name the evidence, expected behavior change, or metric that would indicate success, the proposed test is not ready for prioritization.

Human reviewers still need to decide whether an opportunity fits commercial goals, brand constraints, technical capacity, and the wider customer journey. They must also check whether tracking is reliable and whether a pattern reflects meaningful behavior rather than a reporting issue, a seasonal shift, or an incomplete segment.

Claude's main contribution is therefore faster movement from raw material to a reviewable working brief. It can reduce the manual effort involved in sorting reports and formatting findings. It cannot guarantee that a hypothesis will improve performance or select the right experiment without informed human judgment.

For businesses already using analytics and marketing tools, the most sensible adoption path is incremental: begin with a limited, exported evidence pack; use a repeatable prompt and output structure; compare Claude's triage with an analyst's review; then consider narrowly scoped MCP connections where live access removes a real bottleneck. This approach keeps the process grounded in evidence rather than letting fluent language create unwarranted confidence.

If your team wants to connect AI assistants to analytics or CRM data without creating uncontrolled access, Scalevise can help design read-only, least-privilege MCP integrations that preserve useful audit trails and reduce manual data handling. A well-scoped connection can make recurring analysis faster while keeping people responsible for validation and business decisions. Discuss an MCP setup project with Scalevise to map a practical, controlled path from your existing data sources to AI-assisted workflows.

Frequently Asked Questions

What can Claude do in a CRO audit?

Claude can help triage data, organize evidence, and draft structured findings from supplied materials. It can support tasks such as reviewing a landing-page report and identifying pages or segments that need further validation.

Can Claude validate CRO data or guarantee conversion improvements?

No. The workflow states that Claude cannot determine whether data is trustworthy or guarantee performance improvements. Human reviewers must validate data quality, interpret business constraints, and decide what to test.

When should a team use exported data instead of MCP?

Exported files suit a bounded, point-in-time audit task. MCP connections are intended for controlled live access to approved sources when that access is needed for the workflow.

What safeguards should apply to MCP-connected CRO data?

The documented safeguards include read-only access, least-privilege permissions, data minimization, and an audit trail. Human validation remains necessary when interpreting outputs and setting priorities.

Conclusion

Claude can make CRO audits more efficient when it is assigned a narrow, evidence-led role. Its value lies in accelerating data triage and producing organized material for review, not in replacing the analyst who validates the data and chooses the tests. Teams that define inputs, outputs, access controls, and human checkpoints can use the workflow to save time without confusing AI-generated explanations with proven insight.

ChatGPT Traffic Rose 48% in the US as AI Discovery Reshapes Search Measurement

Ali Farhat — Fri, 11 Sep 2026 20:15:30 +0000

ChatGPT is becoming a more significant discovery surface in the United States. Semrush's Traffic Analytics data for July 2026 estimates that chatgpt.com received about 1.09 billion visits, a 48.38% increase year over year. In the same dataset, bing.com referrals declined about 50.43%, while Google's direct figures rose 10.72%. The numbers do not mean traditional search has been displaced, but they do show that AI-assisted discovery is growing faster than established channels.

For marketers and website owners, the more immediate issue is measurement. Traffic that begins in an AI chat interface can be harder to identify and attribute consistently than a conventional search visit. That makes it risky to judge channel performance using a single analytics view, particularly when businesses are trying to understand whether their content is being discovered through ChatGPT, Google, Bing, or a combination of those paths.

What Semrush's July data shows

The Semrush US trending websites data is based on Traffic Analytics clickstream estimates across desktop and mobile. Semrush says its panel includes roughly 200 million anonymized users. These figures are therefore third-party estimates, not first-party traffic reports from OpenAI, Google, or Microsoft, but they provide a useful directional view of changing traffic patterns.

Surface or metric	July 2026 US signal	Year-over-year change
chatgpt.com visits	Approximately 1.09 billion visits	Up 48.38%
bing.com referrals	Referral activity tracked in the same dataset	Down about 50.43%
Google direct figures	Google remained the largest referrer	Up about 10.72%

The table should not be read as a like-for-like market-share comparison. It combines different signals reported within the same analysis, including visits and referrals. Its value is in the contrast: ChatGPT's growth was substantially faster than Google's, while Bing showed a sizable decline in referrals.

Semrush's wider research adds context. Its April 2026 study reported that outbound referral traffic from ChatGPT grew 206% during 2025. It also found that Google represented a sizable share of ChatGPT referrals. In July, Semrush reported that roughly 21.6% of ChatGPT referrals went to Google, and more than 30% of referrals went to the top 10 domains. That concentration matters because growing AI referral traffic may not be distributed evenly across the web.

AI discovery can still lead back to search

AI-assisted discovery is not necessarily a replacement for search engines. A user may begin with a conversational question, receive a response with links or search prompts, then continue their journey on Google or another destination. Semrush's referral findings suggest that ChatGPT can also send meaningful traffic toward Google.

For businesses, this means the practical question is not simply whether to choose AI platforms or search engines. It is whether useful, trustworthy content can be found and cited across both environments. A clear service page, product explanation, help article, or original research asset may support conventional search visibility while also giving AI systems more material to surface or reference.

Why GA4 attribution needs closer scrutiny

GA4 remains useful for measuring owned web activity, but AI-driven referrals create attribution challenges. Referral data depends on how the originating surface passes traffic to a browser and how that visit is captured. In AI chat interfaces, those pathways may not always appear as a clean, consistent referral source.

That does not make GA4 unusable. It does mean teams should avoid treating a low or missing AI referral count as proof that AI discovery is irrelevant. The underlying visitor may arrive through a link, a copied URL, a follow-up search, or another path that obscures the original interaction.

A more resilient reporting approach is to combine several signals:

Review GA4 acquisition reports for identifiable referral sources and changes in direct traffic.
Compare analytics trends with third-party market data, while recognizing that panel-based estimates are directional.
Track branded search interest, referral patterns, and landing-page performance together rather than relying on one channel label.
Ask customers and leads how they found the business when AI-assisted discovery could be influencing the journey.

Practical priorities for website owners

The July data does not justify abandoning Google-focused SEO. Google remained the largest referrer in Semrush's analysis, and its reported growth was still positive. Instead, it supports a broader content and measurement strategy that does not assume every valuable discovery journey starts with a conventional search result.

Prioritize pages that answer specific customer questions directly, identify the business and its offerings clearly, and provide details that can be verified on the page. Keep important information accessible instead of burying it in image-only assets or vague marketing copy. Then measure whether these pages generate engaged visits, inquiries, and revenue, not merely rankings or referral labels.

AI discovery is becoming another visibility layer that businesses need to understand. Scalevise helps teams assess where their brand appears in AI-generated answers, identify content gaps, and turn those findings into practical website improvements that support qualified discovery. Use the Scalevise AI Visibility and GEO Checker to see how your business shows up across AI search experiences and start an AI visibility scan.

Frequently Asked Questions

How much did ChatGPT traffic grow in the US?

Semrush estimated that chatgpt.com received approximately 1.09 billion US visits in July 2026, up 48.38% year over year.

Did Google lose its position as the largest referrer?

No. Semrush's analysis indicated that Google remained the largest referrer. Its direct figures rose about 10.72% year over year, although that was slower than ChatGPT's reported growth.

Does the Semrush data prove that ChatGPT is replacing search engines?

No. The data shows a strong increase in ChatGPT traffic and changing referral patterns. Semrush also reported that Google receives a sizable share of ChatGPT referrals, indicating that AI-assisted discovery and search can be connected.

Why can AI referral traffic be difficult to measure in GA4?

AI chat surfaces can pass visitors to websites through pathways that are not always captured as a consistent referral source. As a result, analytics reports may not fully show the role AI played in a visitor's discovery journey.

Conclusion

Semrush's July 2026 data offers a clear signal that ChatGPT is growing rapidly as a US discovery surface, even as Google remains the dominant referrer. The priority for businesses is not to declare traditional SEO obsolete. It is to build content that can serve both search and AI-assisted journeys, while using measurement methods that account for incomplete attribution.

Google Search Console’s Missing June Data Was a Reporting Glitch, Not Deindexing

Ali Farhat — Fri, 11 Sep 2026 18:45:30 +0000

Google Search Console experienced a series of reporting problems in June 2026, including a prolonged delay in the Page Indexing report. For many site owners, the report stopped updating on June 11, making it appear that indexing activity had stalled or that pages had disappeared from Google. The evidence later showed a reporting lag, not a broad indexing failure.

Google documented the incident among its official Search Console data anomalies. The Page Indexing data pipeline was subsequently fixed and the report refreshed to show data through June 29. For businesses that use Search Console data in weekly SEO reporting, the incident is a useful reminder that a dashboard trend is not always a change in search performance or index coverage.

What happened to Page Indexing data in June?

The Page Indexing report, previously known as the Coverage report, stopped updating for many users on June 11. By around June 26, the delay had persisted for roughly two weeks. Contemporary reporting by Search Engine Land noted that the report timestamps pointed to a data lag rather than evidence that Google had suddenly removed large numbers of pages from its index.

Independent reporting published on July 3 said Google had repaired the indexing data pipeline and refreshed the report with data through June 29. That recovery clarified the apparent mid-June decline: it was a historical reporting artifact, not a sudden mass deindexing event.

The June issues were not limited to the Page Indexing report. Google's data anomalies documentation also records a June 24 logging error affecting Discover performance data. For properties with AI features in Discover, affected impressions were included in that issue. Google later stated, in an August 21 update, that missing data for the affected period had been restored in Performance reports.

Period or report	What changed	What it meant
Page Indexing report, from June 11	The report stopped updating for many users.	Stale dashboard data could resemble an indexing decline.
Page Indexing recovery, reported July 3	Google fixed the indexing data pipeline and refreshed data through June 29.	The mid-June gap was confirmed as a reporting issue.
Discover performance data, June 24	A logging error affected Discover data and some related AI feature impressions.	Performance reporting for that period was incomplete until restoration.
Performance report update, August 21	Google said missing data for the Discover issue had been restored.	Historical reports needed to be checked again after the backfill.

How to audit SEO dashboards after a reporting disruption

The key operational distinction is between reported data and a site's current status in Google. A delayed report can distort trend lines, alerts, client updates, and internal forecasts even when Google's indexing systems continue to process pages.

Teams reviewing June 2026 data should take a measured approach:

Recheck the current status of important pages with Search Console's URL Inspection tool.
Use site: queries as a corroborating check for important URLs and sections of a site.
Compare the date of an apparent decline with the documented Page Indexing delay before treating it as a technical SEO incident.
Add an annotation to dashboards and recurring reports covering the affected mid-June period.
Revisit reports after Google backfills data, because an earlier export or screenshot may no longer reflect the restored historical record.

This approach helps avoid a costly and distracting response to a false signal. Reworking internal linking, requesting unnecessary reindexing, or escalating a supposed deindexing problem based only on a stale report can consume time without addressing a real issue.

It also strengthens reporting discipline beyond this specific event. Search Console is an important source of first-party search data, but its reports are still data products with logging, processing, and refresh cycles. Where a metric drives a major decision, it is sensible to compare it with direct URL checks and to preserve dated annotations explaining known anomalies.

For agencies and in-house teams, the practical task is to correct the narrative around affected reports. A sharp mid-June movement in Page Indexing data should not be presented as a verified loss of index coverage when the reporting pipeline itself was delayed. Similarly, historical Discover analyses should account for the June 24 logging issue and Google's later restoration notice.

Reliable search reporting matters because teams can otherwise spend time fixing a problem that exists only in a dashboard. Scalevise helps businesses connect visibility data to practical decisions, including checks that distinguish a reporting anomaly from a genuine discoverability issue. Our AI Visibility and GEO Checker provides a structured way to assess how your brand appears across AI-driven search experiences while you keep conventional SEO signals in context. Start an AI Visibility scan.

Frequently Asked Questions

Did Google Search Console stop indexing pages in June 2026?

No. The documented issue was a delay in Page Indexing reporting data for many users. The subsequent data-pipeline fix and refresh indicated that the apparent mid-June decline was a reporting glitch, not mass deindexing.

When did the Page Indexing report stop updating?

For many users, the Page Indexing report stopped updating on June 11, 2026. By around June 26, the report had been delayed for roughly two weeks.

How can I check whether an important page is currently indexed?

Use Search Console's URL Inspection tool to check the current status of an important URL. A site: query can provide a corroborating check while the reporting data is delayed or being refreshed.

Should I change historical SEO reports for June 2026?

Yes. Annotate the affected period and recheck reports after Google's backfill. Do not interpret the mid-June Page Indexing gap as confirmed evidence of lost index coverage.

Was Discover data affected too?

Yes. Google recorded a June 24 logging error affecting Discover performance data and some related AI feature impressions. Google later said missing data for that period had been restored in Performance reports.

Conclusion

The June 2026 Search Console incident demonstrates why SEO reporting should be interpreted alongside direct checks of important pages. Google ultimately restored the delayed Page Indexing data and addressed the separate Discover reporting gap. For affected teams, the appropriate response is to update historical dashboards, verify current page status, and avoid treating the mid-June data gap as evidence of an indexing collapse.

SEOquake Relaunch Adds 28 In-Browser SEO Checks for Faster Page Reviews

Ali Farhat — Fri, 11 Sep 2026 18:30:30 +0000

Semrush has relaunched SEOquake, its free browser extension for page-level SEO checks. The rebuild expands the extension's audit workflow to 28 checks across page, mobile, technical and social signals, while restoring the SEObar experience and redesigning its Quick View and Full Report interfaces. The goal is straightforward: let users inspect a webpage without leaving the tab they are reviewing.

For website owners, editors and in-house marketing teams, the update makes SEOquake more useful as a rapid quality-control tool. It is designed for checking an individual page during publishing, reviewing a competitor page, or spotting visible issues before escalating to a wider technical audit. Semrush positions it as a complement to, not a replacement for, its broader Site Audit toolkit.

What the SEOquake relaunch changes

The official Semrush SEOquake relaunch announcement describes a substantial rewrite rather than a minor interface refresh. The extension now combines familiar at-a-glance information with a deeper reporting workflow, allowing users to move from a quick page review to more detailed analysis in the browser.

The central change is the broader set of page audit checks. Semrush says the rebuilt version covers 28 checks across four areas:

Page signals, which support an immediate review of the page itself.
Mobile signals, which bring mobile-focused checks into the in-browser audit.
Technical signals, for identifying page-level technical SEO issues.
Social signals, which help review information relevant to social sharing.

SEOquake also adds deeper content and link analysis, plus expanded report exports in CSV and PDF formats. Where Semrush metrics are available, the extension can integrate them into the workflow. That makes the tool more practical for a person who needs to record findings, share a page review, or move from a quick observation to a documented follow-up.

Area	Earlier SEOquake experience	Relaunched SEOquake
Page audit scope	Earlier page audit checks	28 checks across page, mobile, technical and social signals
In-browser workflow	SEOquake extension tools	Restored SEObar with redesigned Quick View and Full Report workflows
Analysis and reporting	In-page reports	Deeper content and link analysis, with CSV and PDF exports
Metrics	Included some outdated public-data metrics	Removes metrics such as Google Cache dates and index counts that no longer have reliable public data
Browser support at launch	Not specified in the relaunch comparison	Chrome and Chromium-based browsers including Edge, Brave and Arc. Firefox and Opera are not included.

The removal of older metrics is an important part of the product decision. Google Cache dates and index counts are no longer reliable public data points, so retaining them could create a false sense of precision. The relaunch instead concentrates on information that can support an immediate assessment of the webpage in front of the user.

SEOquake remains a free Chrome extension delivered through the Chrome Web Store. The initial release also works in Chromium-based browsers, including Edge, Brave and Arc. Firefox and Opera are not part of this rollout, which matters for teams that standardize on those browsers.

How SEOquake fits into an SEO workflow

SEOquake is best understood as a rapid page-review layer. It can help a marketer or editor ask practical questions while working: Does this page show obvious on-page, mobile, technical or social issues? Do the content and link details need a closer look? Is there a report worth exporting for the person responsible for fixing the page?

That focus makes it distinct from a sitewide crawler. A page-level extension can speed up checks during content creation, publishing and competitor research because it runs in the same browser context as the page. A comprehensive site audit remains more appropriate when the task is finding issues across many URLs, tracking a larger technical backlog, or reviewing the wider website systematically.

The rebuilt Quick View and Full Report flow supports a sensible sequence: begin with a fast scan, investigate the detail that warrants attention, then export findings if they need to be shared. This can reduce the friction of routine checks, but it does not remove the need to prioritize fixes based on business value and the broader health of the site.

For businesses managing SEO with limited time, the useful discipline is to make page reviews repeatable rather than treat the extension as a one-off diagnostic. Use the same checks for high-value new pages and major updates, document recurring problems, and reserve deeper sitewide analysis for patterns that individual page checks reveal.

Fast SEO checks are useful only when they lead to clearer decisions about visibility and content priorities. Scalevise can help businesses connect day-to-day website improvements with a broader view of how their brand appears in AI-driven discovery. Use the AI Visibility and GEO Checker to identify where your brand is represented in AI answer engines and turn visibility gaps into a focused improvement plan. Start an AI Visibility scan.

Frequently Asked Questions

What is new in the SEOquake relaunch?

Semrush rebuilt SEOquake with a restored SEObar, redesigned Quick View and Full Report workflows, 28 page-level audit checks, deeper content and link analysis, CSV and PDF exports, and Semrush metric integration where available.

Is SEOquake free?

Yes. Semrush describes SEOquake as a free Chrome extension delivered through the Chrome Web Store.

Which browsers support the rebuilt SEOquake?

The initial release supports Chrome and Chromium-based browsers including Edge, Brave and Arc. Firefox and Opera are not included in this rollout.

Can SEOquake replace a full site audit?

No. Semrush positions SEOquake as a quick, lightweight tool for immediate page-level insight. It complements the broader Semrush Site Audit toolkit rather than replacing it.

Why were Google Cache dates and index counts removed?

Semrush removed outdated metrics, including Google Cache dates and index counts, because reliable public data is no longer available for them.

Conclusion

The SEOquake relaunch strengthens a familiar free extension for fast, page-level SEO reviews. Its 28 checks, redesigned reporting flow and exports make it better suited to routine browser-based inspection, while its narrower scope remains appropriately separate from a full sitewide audit. For teams that work mainly in Chrome or another Chromium browser, it provides a more focused way to catch and document issues without interrupting the publishing or research workflow.

2026 SEO Survey: Search Intent Leads, but Relevant Backlinks Still Matter

Ali Farhat — Fri, 11 Sep 2026 00:15:30 +0000

Search intent is the leading priority in a new survey of SEO professionals, with backlinks close behind and content quality in third place. For businesses planning organic search work in 2026, the practical message is not to chase a single ranking lever. Build pages that answer the searcher’s actual need, support them with distinctive information, and earn links from credible sites that are genuinely relevant to the topic.

The findings come from Zyppy's Google Ranking Factors Expert Survey 2026, which surveyed 131 SEO professionals and collected 13,665 data points across more than 100 potential ranking factors. It is important to read the results correctly: this is practitioner sentiment, not an official Google ranking-factor list or a disclosure of how Google’s algorithm works.

Still, the survey offers a useful planning framework. Relevance, or search intent match, was named among the top three factors by 57.1% of respondents. Backlinks followed at 54.8%, while content quality was selected by 47.6%. Search Engine Land’s coverage of the survey reaches the same broad conclusion: intent and relevance lead the list, but links and strong content remain central to practitioners’ SEO priorities.

What the survey says about SEO priorities

Survey priority	Respondents placing it in their top three	Practical planning implication
Relevance, or search intent match	57.1%	Match the page format, depth, and answer to what searchers are trying to accomplish.
Backlinks	54.8%	Prioritize links from trusted, topically relevant domains rather than link volume alone.
Content quality	47.6%	Produce useful, differentiated content, including original research and first-party data where possible.

The close results matter. A useful page with no external signals may struggle to stand out in a competitive market. Conversely, links cannot compensate indefinitely for a page that answers the wrong question or offers little value once a visitor arrives. Treat the three priorities as connected parts of one search strategy, not as isolated checklist items.

Start with the job behind the query

Search intent is the reason someone enters a query. A person searching for a product category may want to compare options. Someone searching for a specific problem may need instructions. A local query may signal that the user wants a provider, location, or appointment rather than a long educational guide.

Before assigning or refreshing a page, teams should inspect the wording of the query and the pages that already rank. The aim is not to copy competitors. It is to establish the likely format and expectation that search results are rewarding, then create a clearer or more useful version.

A practical content brief should define:

the question, problem, or decision the page must address
the intended audience and the stage of their journey
the appropriate format, such as a guide, service page, category page, comparison, or FAQ
the evidence, examples, or product details needed to make the answer useful
the next action that naturally follows from the visitor’s intent

This approach also helps avoid a common failure mode: publishing broad content around a keyword when the searcher wants a specific answer. A page can be well written and still miss the intent behind the query.

Make original information part of content quality

The survey’s content-quality category specifically highlights original research and first-party data. That does not mean every company needs to run a large industry study. The underlying opportunity is to publish information competitors cannot easily duplicate because it comes from direct experience, customers, operations, or proprietary analysis.

For example, a business might turn anonymized patterns from customer questions into a guide, publish a methodology for evaluating a service, analyze its own operational data where appropriate, or document lessons from implementing a process. The material must remain accurate and useful, but firsthand evidence can make a page more distinctive than a generic summary.

Original content also creates a stronger reason for other sites to reference it. That relationship connects the survey’s third-ranked priority with its second-ranked one: useful information can support link earning when it is relevant to people who publish, recommend, or cite resources in the same field.

Build backlinks for relevance and trust

The survey respondents continue to regard backlinks as highly influential. Its write-up emphasizes links from trusted, relevant domains as especially strong signals, while identifying spammy links as negative signals. The business implication is straightforward: a link-building plan should focus on editorial relevance and credibility, not a target number of links purchased or acquired as quickly as possible.

Useful approaches can include contributing expertise to relevant publications, developing citeable original resources, maintaining accurate partner and supplier listings where they are editorially appropriate, and building relationships with organizations that genuinely serve the same audience. Each tactic should produce a link that makes sense to a reader, not merely to a ranking report.

Teams should be especially cautious about tactics that generate low-quality or unrelated links at scale. The survey does not provide a formula for judging every link, but its findings reinforce a durable principle: context and source quality matter.

Measure business outcomes alongside rankings

Rankings can show whether visibility has changed, but they are not the whole result. A page that rises for an irrelevant query may produce little value. Review organic traffic with evidence of whether visitors are completing the action that matches the page’s purpose, such as submitting an inquiry, requesting a quote, making a purchase, or engaging with the next piece of content.

For 2026 planning, connect each important page or content cluster to a measurable outcome. This makes it easier to decide whether an intent-led rewrite, original-data project, or outreach effort is improving the quality of search traffic rather than simply increasing reports and dashboards.

Google search is only one part of how customers discover businesses. Scalevise's AI Visibility and GEO Checker helps teams assess how their brand appears in AI-generated search experiences, then identify practical opportunities to improve the clarity and authority of the information those systems can use. That work can complement an intent-led content strategy and reduce blind spots as discovery habits change. Start an AI Visibility scan to see where your business stands.

Frequently Asked Questions

What did the Zyppy 2026 SEO survey find?

Among 131 surveyed SEO professionals, 57.1% placed relevance or search intent match in their top three ranking factors, followed by backlinks at 54.8% and content quality at 47.6%.

Does the survey prove that these are Google's official ranking factors?

No. The survey reports practitioner views on ranking factors. It is not an official Google list or an algorithm disclosure.

Why does search intent matter for SEO?

Search intent helps determine whether a page answers what a user is trying to accomplish. A page that uses relevant keywords but delivers the wrong format or answer can fail to meet that need.

What type of backlinks should businesses prioritize?

The survey emphasizes links from trusted, relevant domains and identifies spammy links as negative signals. Prioritize links that are credible and make sense in the context of the referring site.

How can a business create original content without a large research budget?

First-party data, anonymized customer patterns, practical methodologies, and documented operational insights can create differentiated content when they are accurate, useful, and appropriate to publish.

Conclusion

Zyppy's 2026 survey does not replace Google guidance, but it provides a clear view of where SEO practitioners are concentrating their efforts. The strongest plan combines pages that meet real search intent, content with distinctive value, and credible backlinks from relevant sources. Measuring the resulting traffic against conversions and other meaningful actions keeps that work tied to business outcomes.